AdvancedLLM evaluation: grading a model with no right answer
Vibes do not scale and BLEU measures nothing you care about. Eval sets, LLM-as-judge without fooling yourself, and the biases that corrupt it.
AI Engineering
Working with LLMs as a systems problem: getting reliable output, retrieving the right context, giving models tools, and knowing whether any of it actually works.
LLMs
Calling models from Python, forcing structured output, choosing between fine-tuning and retrieval, and grading results.
Advanced
AdvancedFine-tuning vs RAG: how to actually choose
Fine-tuning teaches behaviour, RAG supplies facts. A decision framework, the LoRA maths, real costs, and the cheaper ladder to climb first.
IntermediateGet structured JSON out of an LLM — reliably
Asking a model to reply in JSON works until it does not. Learn why tool schemas beat prompt begging, and how to validate every response before it reaches your code.
RAG
Retrieval-augmented generation done properly — chunking, retrieval evaluation, reranking, and the vector index underneath.
AdvancedBuilding RAG properly: chunking, evaluation, reranking
Most RAG fails at retrieval, not generation. Build a golden set, measure recall@k, run chunking experiments, add a reranker when data says so.
AdvancedVector databases under the hood: HNSW and IVF-PQ
ANN search trades accuracy for speed. How HNSW graphs and IVF-PQ quantisation work, which knobs move recall, and how to choose an index.
IntermediateBuild a chatbot that answers from your own PDFs in Python
Retrieval-augmented generation in plain Python: split a PDF into chunks, find the relevant ones, and let Claude answer using only those. No vector database required.
Agents & Tools
The agent loop and its failure modes, plus designing tool interfaces and MCP servers a model can actually use.
AdvancedAgentic AI architecture: the loop and what breaks
An agent is a loop with tools and a stopping condition. Context growth, error handling, termination — and when to write a chain instead.
AdvancedMCP and tool calling: designing tools an LLM can use
Tool calling fails at the interface, not the model. Schema design, descriptions, error contracts, and what MCP actually standardises.
Newsletter
Get new articles
Free essays on learning, Python, and ML — no account required. We’ll only email when there’s something worth reading.